{"id":"W7125598063","doi":"10.1109/cascon66301.2025.00045","title":"An Empirical Evaluation of LLM-Based Approaches for Code Vulnerability Detection: RAG, SFT, and Dual-Agent Systems","year":2025,"lang":"","type":"article","venue":"","topic":"Software Engineering Research","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"ca_institutions":"Ontario Tech University","funders":"","keywords":"Transparency (behavior); Vulnerability (computing); Audit; Baseline (sea); Domain (mathematical analysis); Software; Task (project management); The Internet","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01483203,0.001786575,0.0009904442,0.006236621,0.0008842111,0.001647727,0.002579483,0.00217513,0.001214402],"category_scores_gemma":[0.04701111,0.0004710242,0.001179454,0.002621689,0.001416576,0.004048508,0.003848963,0.002006467,0.00100038],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001922821,"about_ca_system_score_gemma":0.001994472,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00665756,"about_ca_topic_score_gemma":0.007853733,"domain_scores_codex":[0.9864417,0.006678746,0.001159739,0.002616251,0.002625249,0.0004783937],"domain_scores_gemma":[0.9553109,0.03095455,0.002547191,0.006802679,0.003320971,0.001063682],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.003631232,0.003287518,0.1276944,0.003120471,0.001350977,0.0004717375,0.002292766,0.1385013,0.01008279,0.002999798,0.02308951,0.6834776],"study_design_scores_gemma":[0.0002832551,0.001910238,0.02970835,0.0002415298,0.0003617253,0.0005938048,0.0009280176,0.9386735,0.01475746,0.004079346,0.008329694,0.0001329934],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8843455,0.005450674,0.08400428,0.001059774,0.000246816,0.0008644072,0.003675642,0.01476212,0.005590802],"genre_scores_gemma":[0.8687012,0.0005937717,0.1205806,0.0002311568,0.00007739344,0.0003382883,0.00755222,0.0004342664,0.001491128],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.01483203,"threshold_uncertainty_score":0.07844013,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1558992302502042,"score_gpt":0.3882194443479205,"score_spread":0.2323202140977163,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}